Computer analysis misses the point.

What I generally miss is that chess engines often go for short-term wins and don't take into consideration what the player eventually built up towards. Positional play as a human is different from computer play. Yes, computers play perfectly, but humans also anticipate moves that others may forget, which might result from time pressure or oversight. This is especially true when setting up complex forks - the engines don't understand that humans make preparatory moves for these tactics. When i played a NM for example i learned about a null move in chess, never had seen that before. Computers probably won't do that.
And well if your a 1200 or so, chess is different as compared to a 1800 or 800.
As a human you think about your opponents moves and what he might be aware off.
Some suggestion :

• • "You played a less optimal move (as you could have [whatever seems a short-term win]), however, later in your game this turned out to be a nice choice because it became a valuable part of your main attack/outpost."
  • "While not the best move, did you anticipate your opponent would forget about its implications? You indeed got lucky as they forgot about defending their [piece], and now you've taken it. However, it was risky because they could have also played [alternative moves]."
  • "Your line in the opening isn't a main line of the [specific] opening. Though several grandmasters have played it in certain situations, it's rarely seen against your opponent's opening choice."
  • "You didn't go for that trade that would have eventually won you a pawn, but maybe this fits your game style of building up towards the middle game."
  • "Commonly it's better to take pawns toward the center, but in your case, looking at your opponent's position, it was wise not to do so."
  • "This move didn't do much but your end game depended on it, i think you had a tactical plan.
  • It might have been a tactical plan, however pushing like that is risky eventually you loose that peace, and you lost tempo here.
• Maybe use some historic chess game descriptions train a lllm on it, to see how games are written about by humans

- Games with opening sacrifices are common by humans, but chess engines always count it as a tactical error, while in reality it can brake opening ideas, and surprice.

Computer evaluations are meant to be objective, up to the limits of the engines. A blunder is still a blunder even if your opponent doesn't take advantage of it, misses the correct continuation a few moves later, of has their own blunder later.

There is room in some positions where the human ideas are easier to calculate than the engine's better line, but a lot of the time that is not going to be the case.